Search insights from the AOL data dump

A few weeks ago, AOL released the “anonymous” record of searches made between March and May of this year by 650,000 of its customers.

I put “anonymous” in quotes because it’s possible to identify some individuals by analyzing their search patterns.  Big scandal, so the data was pulled from public view, but before it was…

One analyst, Lee Gomes of the Wall Street Journal, plucked out some gems of information.

Of the 17.15 million searches in the database, here were the most searched words in order from the most to the least:

1. free
2. new
3. lyrics
4. county
5. school
6. city
7. home
8. state
9. pictures
10. music
11. sale
12 beach
13. high
14. map
15. center
16. sex
17. google

What jumps out at me is how many searches seem to be about things that are close by: county, school, city, home, state.  Oh,  the opportunities in creating local portals!

I’ve been singing this song since at least 1995, but few people have taken me up on it  – yet I’ve never heard a complaint from those who have and who have done it with some intelligence.

The power of “free” and “new.” Somehow, more than a century before search, mail order copywriters had already figured out that these two words were hyper-magnetic.

Of the 17 million odd searches in the sample, Gomes says that over 400,000 were in the form of a question using one of the five W’s: who what when where why. My question is how could he forget “how?” If you’re buying keyword phrases, you definitely want to research all the relevant “how to” related phrases.  There’s often gold there.

Here’s some important data: Of all the searches in this sample, 42% of the time people who clicked, clicked on the first link presented to them.  That means that the other 1,000, 10,000, 1,000,000 links – depending on the keyword – had to fight it out over the remaining 56%. Actually, it’s probably worse than that because I bet the top ten links suck up 90% or more of all clicks.

Then there are the 47% of searchers who don’t click on any of the results.  That’s right. They do a search, look at the results and click on nothing.  This may have more to do with the quality of AOL’s search results than anything else. Recalling a survey I saw in the last year. AOL customers aren’t thrillied with AOL search.

Overall, nothing earth-shattering in these results, but it’s always nice to browse through 2.27 gigbytes of search data when the chance presents itself. Thanks AOL.

– Ken McCarthy

P.S. For over 25 years I’ve been sharing the simple but powerful things that matter in business with my clients.

If you’d like direction for your business that will work today, tomorrow and twenty years from now, visit us at the System Club.

Domain name madness
Google Video gets aggressive

21 Responses to Search insights from the AOL data dump

  1. Matt Glenville August 21, 2006 at 10:18 am #

    Some more useful analysis:

    Clickthroughs by position:

    The top position gets 42.3% of all clickthroughs.
    No. 2 gets 11.92%
    No. 3 gets 8.44%

    So the first 3 places get 62.66% of all clickthroughs. (Seems to back up the ‘Google Golden Triangle’ finding – eyetracking experiments show that the top 3 positions are by far the most looked at.)

    By the time you get to posisition No. 11 you’re down to a 0.66% clickthrough rate.

    Clickthroughs by Page:

    Page 1 of the search results gets 89.69% of all clickthroughs.
    Page 2 gets just 4.4%.

    And it drops off rapidly after that.

    You can see the graphs and figures here:

    http://www.redcardinal.ie/search-engine-optimisation/12-08-2006/clickthrough-analysis-of-aol-datatgz/

    This site also has some useful analysis – like a very high percentalge of users type in the full address into the search box, e.g. ‘www.kenmccarthy.com’, rather than ‘ken mccarthy’:

    http://www.seo-portal.com/

    One random, intersting and not really useful finding:

    If your site is below position no. 250 you would actually get more traffic if you were last (position 500)!

    Matt

  2. Ken McCarthy August 21, 2006 at 11:26 pm #

    Matt sent me this interesting picture which shows how people’s eyes track when they look at a Google results page.

    Red is where eyes spend a lot of time, orange less than that, yellow less than that all the way down to dark blue.

    This confirms that the web sites listed at the very top top of the page get the lion’s share of the attention.

    Thanks Matthew.

    Click the link to see the picture.

    http://kenmccarthy.blogs.com/photos/ken_blog/eyetracking.jpg

  3. Ken Frazza August 24, 2006 at 10:18 am #

    Ken,

    A question regarding the AOL clickthroughs by position. Since I never have used AOL search, I went and tried it. The top 3 results were for sponsored links and then the next 10 under them were the (free search) web results.
    So when they say that the 1st position gets 43% of the clicks, are they talking about the 1st sponsored results or the 1st web results(free)?

    Since AOL uses google for it search results, does it still use google for it’s sponsored links? If so, it makes (CTR) Click Thru Rate even more important since you can not just buy your way to the #1 position at Google.

    Thanks,

    KF

  4. tony August 24, 2006 at 10:37 am #

    Hey Ken,

    Long time listener, first time caller 😉

    The AOL “Data Valdez” has a lot of interesting implications — and yuo’re right, some of them have to do with marketing.

    If I might share a few interesting links with your audience?

    Analyzing the raw data for CTR and so on is a wonderful idea.

    Some researchers, however, are having an ethical conundrum:
    http://www.nytimes.com/2006/08/23/technology/23search.html?ex=1313985600&en=cc878412ed34dad0&ei=5088&partner=rssnyt&emc=rss

    — certainly one of the chestnuts here is that people search for their own name; you can’t tell if the names of people coming up on searches correlates with the “anonymous” ID.

    If you’re thinking broading about “marketing”, its a PR nightmare for AOL. Jason Calacanis who sold his blog network for millions now works for AOL (and is no mean marketer himself …)
    http://www.calacanis.com/2006/08/09/hard-times-at-aol-right-now/

    And of course, if you want to do some “mining” yourself, you can hit up one enterprising gentleman who has set up a search engine for the data.
    http://www.aolsearchdatabase.com/

    Love the blog Ken … keep up the great work.

    Cheers
    Tony.

  5. Peter Dudek August 24, 2006 at 11:36 am #

    Ditto the question by Ken Frazza. That was the first thing I wanted to know. And it’s VERY important!

  6. Burnie Smith August 24, 2006 at 1:22 pm #

    Hi Ken,
    Most of my searches are for specific companies or sites where I do not quite remember the URL. My search is specific enough so that most of the time (80% to 90%)my answer is the first choice.

  7. chuck rock August 24, 2006 at 1:55 pm #

    Thanks Ken!!!

  8. Jerry Dyas August 24, 2006 at 2:05 pm #

    Let me ditto the ditto. You hear all the time that you don’t have to be in the top 3, that you want to move down the list. So this data seems to contradict that datum.

  9. Laurie Kristensen August 24, 2006 at 3:11 pm #

    In answer to the question about whether the top 3 position are actually the paid results or the natural results, as a coincidence I opened an email immediately before the one that led me to this page, that offered a free video that exactly answers this question — it’s extremely interesting, and you don’t have to opt in or anything to see it.

    http://www.convertlinks.com/decoded

    The short answer is, it’s the natural results that apply.

    There are more free videos to come that you can also sign up for at the end.

    ~ Laurie Kristensen

  10. Richard Hearne August 24, 2006 at 3:59 pm #

    Hi Ken

    Just a quick note:

    “Then there are the 47% of searchers who don’t click on any of the results. That’s right. They do a search, look at the results and click on nothing.”

    This isn’t quite true. Yes they click on nothing, but many of the people refine their search query to produce better results. So when you look at the dataset you can see the iterative approach people take toward building their search phrases.

    It’s probably more appropriate to say that 47% of people didn’t find what they were looking for!

    Oh, and just to mention, AOL search is powered by Google so, in effect, this dataset is based in Google’s index.

  11. Ken McCarthy August 24, 2006 at 4:03 pm #

    Richard – Thanks for that clarification. Makes sense.

  12. Chris Bennett August 24, 2006 at 4:10 pm #

    Awesome look into the untapped world of Regional and Local Search it is the future of search as we know it for businesses.

  13. Richard Hearne August 24, 2006 at 5:17 pm #

    @Chris Bennett

    Your comment is very interesting for me. I am based in Ireland and all I see is regional and local search. Google already geo-targets it’s results so I see defferent results than someone searching in the US.

    So when I search for widgets on Google.com (or Google.ie) I am presented with results that are biased toward Irish websites (either hosted in Ireland or using the .ie ccTLD).

    Your observation is very true – in the future results might/could become more biased toward the street/town you are searching from. Of course the very concerning downside to this will be the issue of privacy – a balance will have to be struck between user convenience and the tracking possibilities inherent in local search.

  14. Jonathan Gunson August 24, 2006 at 7:01 pm #

    Ken

    You are on the money with ‘local’ search, and also local portals when it comes to getting traffic.

    It gets me HUGE amounts of traffic. (Seriously) I was very surprised at this.

    Here’s how:

    I used to exclusively go WIDE, but now aim articles and PR at LOCAL communities.

    This almost doubles the visitors to my website.

    Another hot subject word you mention is the word FREE. It sure works! I used to sell my book on the ‘viral marketing’ secret the reveals how I got a MILLION hits in 27 days to my website.

    That book made me money, but I took a risk, and now I give it away for free in return for folks joining my mailing lists.

    http://viralalchemy.com

    That was a great decision, it attracts a huge number of interested folks to my list, who then buy other products at a leisurely pace while I give them more great information – again for free.

    But the big point I have discovered is that the LOCAL articles I write offering the free eBook attract far more traffic than the WIDE articles.

    ‘Local’ works.

    Jonathan Gunson

  15. Peter Dudek August 25, 2006 at 12:55 am #

    To Andy and Brad with the Going Natural video at http://www.convertlinks.com...

    I really enjoyed your video. And of course I want to know more. But on the topic of where the eyes go on the page concerning the paid vs. natural listings: the infra-red images from the screen shot you show seem to be mostly focused on the very top left of the page which are where the top three PAID listings are usually found. The screen shot you show only has paid listings on the far right column.

    Are there any infra-red images where the top three listings are paid listings?

    Thanks,

    Peter Dudek
    peterdudek@comcast.net

    P.S. I’m looking forward to your next video and I’ll be interested to find out who you are and where all this is leading. Please consider me a prime prospect for whatever it is you’ll be selling.

  16. SEO Portal August 25, 2006 at 2:33 am #

    I don’t know what sample he used, but there are over 30 milion searches in the AOL data.
    And about the words; I assume these are the single words pulled out of the search queries.. not real searches themselfes. Im not sure what good that info is.

    To me it’s far more interesting to look at the actual searches and see what people search the most:

    http://www.seo-portal.com/aol-data-analysis-iv-where-do-people-search-for/2006/08/15/

  17. Ken McCarthy August 25, 2006 at 8:39 am #

    Peter,

    Good point! I noticed that too. When there are AdWords ads on the top, they get top attention from me – and it looks like the eye tracking

    If the comparison is to the ads on the side of the page, I can see them getting much less attention in contrast with the organic results.

    I’ll pop a note to Brad and maybe he’ll weigh in on this point.

    SEO-Portal: Interesting point about the single word queries. You’re right, of course. I’m sure not many people search just for the word “free.”

    I think the value of this particular aspect of the data is to see themes, for example that the word “free” appears in lots of searches, that people often include specific geographic info in their searchers (ex. shoemarkers in Budapest.)

  18. Ken Frazza August 27, 2006 at 11:35 am #

    Ken,
    Some more questions regarding the “sponsored links” verses “web results” data. Though I do see value in the data, it creates a lot more questions that I hope can be answered.
    We are told that the #1 position gets 43% of the clicks. Since the sponsored links come up on top of an AOL search, isn’t that actually the #1 position? That would mean that the #1 spot on Web results is actually the #4 position. I understand that the data is only on the web results. Was their an analysis done on the sponsored links? If so, was their one done on both combined? Did AOL release this data or does Google handle the data on the sponsored search?

    Did the top 3 sponsored links get more clicks then the top 3 web results?
    Does the average AOL searcher even know the difference between a Sponsored link and web results?

    As someone who has many PPC accounts for myself and some of my customers, any further information would be helpful. I have a lot more questions that I could ask about this.

    I am grateful for you and Richard Hearne for providing this info. I was hoping you could shead some more light on this as the more you know the more you want to know.

    Thanks,

    Ken Frazza

  19. Richard Hearne August 27, 2006 at 12:14 pm #

    @Ken McCarthy – there must be some Irish blood in you with a name like that 🙂

    @Ken Frazza – The dataset contains only organic listings – no sponsered listings. So when someone clicks on the #1 result this means the #1 organic listing.

    No data was released about the sponsered listings or the clickthrough rates for same, but if memory serves me correctly I seem to recall some research into organic vs. sponsered and the majority of those surveyed showed a far higher ‘trust’ in the organic listings. Don’t quote me!

    The point you make about some searchers not knowing the difference between the sponsered and organic listings is very valid. However, it is not in the interests of the major SE’s to try to ‘blend’ in the PPC’s with the organic listings as this would dillute the quality of their main product – those same organic listings. Of course, that’s not to say that this hasn’t been happening over the past few years…

  20. isabel August 27, 2006 at 11:29 pm #

    Very, very, very, interesting, and useful on top. Gratulations. Isabel

  21. lrgotrlsji June 18, 2007 at 5:47 pm #

    Hello! Good Site! Thanks you! pexblacrpsyf

Leave a Reply